Unexpected pgbench result - Mailing list pgsql-performance

From Dave Johansen
Subject Unexpected pgbench result
Date
Msg-id CAAcYxUe11Qdx7h3RaRjuGGvp3oO7Wi-QNYrN=9TXBSv+P2u0hg@mail.gmail.com
Whole thread Raw
Responses Re: Unexpected pgbench result  (Shaun Thomas <sthomas@optionshouse.com>)
List pgsql-performance
I'm working on setting up a large database (or at least what I consider to be a large one with several tables having 10-20 million records inserted per day), and I've been using pgbench to verify that the hardware and database are configured in an optimal manner.

When I run pgbench in "SELECT only" after doing "-i -s 2000" I get what appears to be good performance (60k-70k tps) but if I initialize a new database with "-i -s 4000" the tps drops to 4k-7k. Is this order of magnitude drop expected? Or is there something wrong with my hardware or database configuration that is causing this issue?

The one option that I've been considering is partitioning and I've been asking about it here:
http://www.postgresql.org/message-id/CAAcYxUcb0NFfMDsMOCL5scNRrUL7=9hKxjz021JMQp0r7f5sCQ@mail.gmail.com

pgsql-performance by date:

Previous
From: Dave Johansen
Date:
Subject: Re: Recommendations for partitioning?
Next
From: Scott Marlowe
Date:
Subject: Re: Recommendations for partitioning?